Faculty Advisor:
Brad A. Myers, Senior Research Scientist, Human Computer Interaction Institute, School of Computer Science

Grant:
Funded $19,800 on 9/1/00
Funded $18,300 on 7/1/99
Funded $17,700 on 7/1/98

Project:
Lightweight Structured Text Processing
Examples of structured text abound -- Web pages, email messages, bibliographies, address lists, source code. In fact, almost all text has some structure. Text editors recognize this fact to some extent by providing operations on characters, words, lines, and paragraphs, but extending this set to include user-defined objects like titles, authors, university course numbers, and variable identifiers is not an easy task. Lightweight Structured Text Editing (LSTE) solves this problem by enabling the user to define and manipulate text structure interactively. With an LSTE system, users can search, filter, sort, edit, and reformat semi-structured documents and document collections. Key contributions include a pattern language that is simple and usable, robust pattern matching that degrades gracefully in the face of irregular syntax, and a graphical interface that helps the user discover and correct mismatches.

Current status: September 2001
LAPIS has been extended with two powerful new features:

• Simultaneous editing, a technique for automating repetitive text edits interactively. Simultaneous editing can do virtually anything that a recorded keyboard macro can, but lightweight structure knowledge makes it far easier to edit structured text like HTML or Java source code. More details can be found in our USENIX 2001 paper, Interactive Simultaneous Editing of Multiple Text Regions.
• Outlier finding, which helps find bugs in text-matching patterns by bringing unusual matches to the user's attention. More details are in an upcoming UIST '01 paper, Outlier Finding: Focusing User Attention on Possible Errors